Access to this full-text is provided by Springer Nature.
Content available from Scientific Reports
This content is subject to copyright. Terms and conditions apply.
Scientic Reports | (2021) 11:8507 |
www.nature.com/scientificreports
The other‑race eect and holistic
processing across racial groups
Hoo Keat Wong1*, Alejandro J. Estudillo1,2, Ian D. Stephen3,4 & David R. T. Keeble1
It is widely accepted that holistic processing is important for face perception. However, it remains
unclear whether the other‑race eect (ORE) (i.e. superior recognition for own‑race faces) arises from
reduced holistic processing of other‑race faces. To address this issue, we adopted a cross‑cultural
design where Malaysian Chinese, African, European Caucasian and Australian Caucasian participants
performed four dierent tasks: (1) yes–no face recognition, (2) composite, (3) whole‑part and (4)
global–local tasks. Each face task was completed with unfamiliar own‑ and other‑race faces. Results
showed a pronounced ORE in the face recognition task. Both composite‑face and whole‑part eects
were found; however, these holistic eects did not appear to be stronger for other‑race faces than for
own‑race faces. In the global–local task, Malaysian Chinese and African participants demonstrated a
stronger global processing bias compared to both European‑ and Australian‑Caucasian participants.
Importantly, we found little or no cross‑task correlation between any of the holistic processing
measures and face recognition ability. Overall, our ndings cast doubt on the prevailing account that
the ORE in face recognition is due to reduced holistic processing in other‑race faces. Further studies
should adopt an interactionist approach taking into account cultural, motivational, and socio‑
cognitive factors.
e other-race eect (ORE; also known as the own-race bias) is a well-documented phenomenon showing that
people are generally better at recognizing faces of their own race, compared to faces of dierent races. It exists
across dierent countries and ethnic groups1 and is evident not only in laboratory settings but also in real-world
scenarios2. Although the ORE has been extensively studied for the last four decades, the specic mechanisms
underlying this eect are still poorly understood. e present paper aims to shed light on this issue by exploring
the holistic processing account of the ORE3.
According to a long-standing scientic tradition, holistic processing is the hallmark of adults’ expert face
recognition4. While the exact denition of holistic processing is a matter of ongoing debate, it is widely accepted
that when adults perceive faces holistically, the facial components (e.g., eyes, nose, mouth) are integrated into
a whole or gestalt-like representation4,5. Two experimental paradigms have been widely employed as standard
measures of face-specic holistic processing: the whole-part task and the composite face task. In the whole-part
task6,7, recognition memory of a facial part (e.g., the eyes) is more accurate when it is presented in the context
of a whole face than in isolation, suggesting that facial features are embedded into a holistic face percept. In the
composite face task4, observers’ performance on matching two identical top face halves is better when these
top halves are misaligned (i.e., spatially oset) with dierent bottom halves than when the top and the bottom
parts are aligned. is composite eect demonstrates that the face parts are not perceived independently from
the whole face.
Holistic processing has been proposed as one important mechanism underlying the ORE. According to this
view, in contrast to own-race faces, people are inecient at integrating facial components from other races into
a whole representation8,9, and therefore other-race faces might be subject to weaker holistic processing than
own-race faces. Although a stronger holistic processing for own-race faces compared to other race faces has been
reported using the whole-part task9 and the composite face task10, these results are not always replicated11–13.
In fact, the results obtained from the composite task are very inconsistent8,11,14, and certainly not as consistent
as those from the whole-part task. e discrepancy in the holistic eect results may stem from methodologi-
cal dierences between studies (e.g., face size15, measuring methods10,16, limited construct validity of holistic
processing17–19, and independent sample collection from race groups who have dierential level of interracial
experience9,10,12). Yet, these observations lend support to the claim that the holistic mode of processing faces
allows ecient encoding of an individual face20 and can be moderated by the race of observer21.
OPEN
School of Psychology, University of Nottingham Malaysia, Semenyih, Malaysia. Department of Psychology,
Bournemouth University, Dorset, UK. Department of Psychology, Macquarie University, Macquarie Park,
Australia. Perception in Action Research Centre, Macquarie University, Macquarie Park, Australia. *email:
hookeat.wong@nottingham.edu.my
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Vol:.(1234567890)
Scientic Reports | (2021) 11:8507 |
www.nature.com/scientificreports/
Limited experience with other-race faces has been proposed as one of the causes of the reduced holistic
processing for other-race faces, and therefore the robust ORE. For example, in the aforementioned studies,
Caucasian observers had very limited exposure to Asian faces in either daily life or the media; in contrast, Asian
participants in the these studies were international students in Western universities and reported having similar
amount of social contact with own-race and other-race individuals8,22. Yet, this experience-based explanation of
holistic processing has been questioned because other studies have found equivalent levels of holistic processing
for both own- and other-race faces in Asian participants with limited exposure to other-race faces10,12,13,23,24.
An explanation for the roughly equivalent holistic processing magnitude for own-race and other-race faces
found in the Asian samples is that, compared to Caucasians, Asians are more prone to holistic processing of
both face and non-face visual stimuli. For example, Asian observers exhibit a stronger global processing bias in
the classical Navon task than Caucasian observers25. Not only does this theoretical explanation underline the
cultural dierences in cognitive styles between Caucasians and Asians, but it also implies that holistic process-
ing detected for other-race faces in Asian participants may be attributable to domain-general global processing
bias instead of specialised higher-level mechanisms for face recognition, as argued by Michel etal.10,26. Based on
such a general cognitive style, Asians may maintain a relatively broad facial representation that is advantageous
for recognising both own- and other-race faces, thereby reducing the ORE. is may further explain why some
researchers failed to observe the ORE in Asian samples27,28. Although empirical studies have set out to explore
the association between domain-general global processes, face recognition ability, and face-specic holistic
processing29,30, only a few studies directly evaluated its validity by comparing between multiple ethnic groups
with the use of face stimuli of dierent races. For instance, DeGutis etal.’s16 and Wang etal.’s31 conclusion that
recognition ability is strongly linked to the magnitude of holistic processing lack external validity as the former
study only tested a Caucasian participant sample with the use of Caucasian faces, whereas the latter study did
not report the race of participants and only used Asian face stimuli.
The present study. e widespread assumption in the face perception literature is that the whole-part and
the composite face tasks measure the same underlying (holistic) mechanisms32–36. However, a recent study found
no association between these two tasks37, suggesting that they, in fact, tap dierent perceptual mechanisms. So
far, only one recent study13 employed both composite-face and whole-part tasks to index holistic processing
while comparing between two dierent race groups (Caucasian vs. Chinese). Mondloch etal. reported evidence
that the magnitude of holistic processing for own-race and other-race faces did not dier in both Caucasian and
Chinese adults. However, this cross-racial study did not measure participants’ face recognition memory and
therefore it remains unclear to what extent holistic processing aects the ORE in recognition memory.
In the present study, we investigate whether the ORE in face memory can be attributed to reduced holistic
processing (as indexed by both composite-face and whole-part eects) of unfamiliar other-race faces. To increase
the generalizability of our results, we test face recognition ability and holistic processing in Malaysian Chinese,
African, European Caucasian, and Australian Caucasian young adults using three races of faces (Chinese, Cau-
casian and African faces). If holistic processing is important for recognising faces and individual-level face
discrimination experience is crucial for holistic processing to develop, we would expect that participants from
dierent race groups will show the typical ORE in face memory, and stronger holistic processing for own-race
faces than other-race faces. Alternatively, if holistic processing can be generalised to facial morphologies that
are less visually experienced without extensive individuating (e.g.38–40), both own- and other-race faces would
elicit holistic eects of similar magnitudes across race groups.
In addition, we used Navon gures to compare global–local processing dierences between the four race
groups. Based on the accumulated evidence of stronger global processing but weaker local processing in East
Asians compared to Western Caucasians41, we predicted that Malaysian Chinese would be more susceptible to
global–local interference (GLI)—an index of the tendency to globally process general objects—than Caucasian
groups (European and Australian). Such a perceptual dierence indicates that information-gathering strategy
(global versus local processing) for general stimuli can be culture-dependent25,42, with collectivist societies (i.e.,
the East) producing a preference for integrating context, and individualist societies (i.e., the West) producing
a preference for ignoring context43. Like South-East Asia, African cultures are also considered collectivistic44,
but research on cultural dierences in perceptual processing bias has oen neglected this population. To ensure
valid theoretical conclusions, we also tested African participants from collectivistic societies and hypothesised
that they would show an evident GLI (i.e. faster and more accurate at global processing).
Furthermore, if the mechanisms involved in holistic processing can apply to other object classes (e.g. Navon
letters) and are not specialised for faces per se (“domain-generality hypothesis”), then GLI scores would vary
systematically with performance on both the whole-part task and the composite face task. Conversely, if special
mechanisms are involved in processing faces holistically (“domain specicity hypothesis”), the magnitude of GLI
would not correlate with holistic face processing measures and face recognition ability, such that perceptual biases
for general information processing is not necessarily generalisable to high-level, specialised face processing.
Method
Participants. irty-one Malaysian-Chinese (16 females; Mage = 21.65, SD = 2.6), 30 European Caucasians
(14 females; Mage = 22.40, SD = 3.10), 30 Australian Caucasians (23 females; Mage = 21.03, SD = 4.45), and 30 Afri-
cans (12 females; Mage = 26, SD = 5.5) took part in this study. All participants self-reported single rather than
mixed-race descent. Malaysian Chinese were students studying at the University of Nottingham Malaysia. ey
were all born and grew up in Malaysia. None of them reported spending more than 9months outside Malaysia.
European Caucasian and African participants were international students recruited at the University of Notting-
ham Malaysia. European-Caucasians were mostly British (one Italian, one Dutch) who had resided in Malaysia
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Vol.:(0123456789)
Scientic Reports | (2021) 11:8507 |
www.nature.com/scientificreports/
for 6.5months on average. None reported spending more than 2years in a predominantly Asian country. Afri-
can participants were mostly Nigerians (ve Kenyans, two Zimbabweans, one Zambian, one Somali) who had
resided in Malaysia for 1.5years on average. Australian-Caucasian participants were recruited from Macquarie
University, Sydney. All were born in Australia and had not lived in a predominantly Asian country for more than
4months (M = 5.4 days, SD = 21, range 0–120days). All participants reported having normal or corrected-to-
normal vision and having no diculty with face recognition. All experimental protocols were approved by the
University of Nottingham Malaysia, Faculty of Science Ethics Committee, and all methods were carried out in
accordance with guidelines of the British Psychological Society. e individuals depicted in all gures signed a
written informed consent for their images to be published. Participants gave written informed consent prior to
the experiment and received either course credit or monetary compensation of RM10 (approximately US$3) for
their participation.
A priori power analysis using G*Power 3.1.9.245 showed that, for all of the terms in our analyses that directly
related to our hypotheses (all of which are 4 × 3 within-between interactions in mixed ANOVAs), this sample
size gave sucient power to detect eect sizes of ηp2 < 0.06 (a small-medium eect size), with α = 0.05, and power
(1 − β) = 0.80.
Apparatus, stimuli and procedure. Chinese, Caucasian and African faces were used. Chinese facial
images were collected from a student population at the University of Nottingham Malaysia Campus; Caucasian
faces were obtained from students at Macquarie University, Australia. African faces were requested from Coet-
zee’s46 face database. All stimuli used in the face tasks were frontal images of young adult faces (both male and
female) with neutral expression, and no glasses, facial hair, or distinctive blemishes (see Fig.1). Individual face
identities did not appear in more than one task. Considering that face photograph memorability is inuenced
by a combination of facial properties such as distinctiveness and attractiveness47, 216 face images (72 for each
race) were originally sampled according to the results of a prior experiment in which each face race was matched
in terms of attractiveness and distinctiveness as rated by 95 young adult participants (24 Chinese, 24 Malay, 25
Indian, and 21 Caucasian) on a 7-point Likert scale48. is selection criterion minimised potential confounds of
facial distinctiveness and attractiveness on participants’ recognition performance. e original images were rst
cropped to form an ellipse shape that excluded external features (leaving a roughly oval shape with no hair on the
top and sides). To minimise the low-level image cues (e.g., skin colour information), all face images were trans-
formed into 8-bit grayscale images in Adobe Photoshop CS6 and were aligned on the eyes’ position using Psy-
chomorph soware49 (http:// users. aber. ac. uk/ bpt/ jpsyc homor ph/, Version 6). Stimuli were presented on a 15.6″
monitor (resolution 1366 × 768). Participants were tested individually in a quiet dimly lit room with three face
tasks (yes–no recognition task, composite task, and whole-part task), in counterbalanced order. Participants also
performed a global–local task; however, as this task induces holistic or featural processing biases50, it was always
performed last. Participants completed all tasks in approximately one hour, including breaks between each task.
Yes–no recognition task. Sixteen faces of each race group (eight females) were selected to form the experimental
set. Each face was presented only once on a light grey background and sized 7.5° horizontal by 10.5° vertical at
approximate viewing distance of 60cm. During the learning phase, participants were asked to passively view and
learn 24 faces (eight per race group). On each trial, a face was presented randomly in one of the four quadrants
for 5 s, preceded by a central xation cross for 1 s. In the recognition phase, 24 learned faces were randomly
intermixed with 24 novel faces. For learned faces, the facial expression (neutral or smiling) changed between
the learning and recognition phases to avoid a trivial image matching strategy. On each trial, participants were
required to indicate as quickly and as accurately as possible whether they had seen the face in the learning phase.
e face was presented for up to 5 s and no trial-by-trial feedback was given. If participants did not respond
within the rst 5s, a blank screen would appear until they responded. Both response times and accuracy were
recorded. Faces were presented in a random order, with the constraint that no more than three trials involving a
given race occurred in immediate succession. e experimental procedure is illustrated in Fig.2.
Whole‑part task. Stimuli were created from 36 face images: 12 target faces (two of each race and sex) and 24
distractor faces containing four faces of each race and sex. Within each race and sex category, a standard face
outline template was used, and each target face was created by aligning eyes, nose, and mouth features into the
template. Distractor faces for the whole trials were created by replacing one feature (i.e., eyes, nose, or mouth)
in the target face with the respective feature of another face of the same race and sex. Part stimuli were cre-
Figure1. Examples of the three races of faces (i.e. Chinese, African and Caucasian faces) used in the face tasks.
Each race pair shows a female (le) and a male (right) face. e individuals depicted in this gure signed a
written informed consent to the publication of their facial images.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Vol:.(1234567890)
Scientic Reports | (2021) 11:8507 |
www.nature.com/scientificreports/
ated by extracting the eye, nose, or mouth region from each of the target faces and the distractor faces. Target
and distractor stimuli for the part trials displayed only the critical feature (see Fig.3). At a viewing distance of
approximately 60cm, whole faces were of 7.5° horizontal by 10.5° vertical and for isolated features the sizes were:
eyes 6.5° × 2.2°; nose 2.6° × 2.2°; mouth 3.8° × 1.9°.
e task comprised three study-test race-blocks (Chinese, Caucasian and African faces). During the study
phase, participants were instructed to memorise four faces (two males) and their associated names (e.g., John,
James, Jill, and Jane). Each face-name pair was shown for 5 s with an inter-stimulus interval of 1 s. Participants
entered the test phase only when they could correctly identify every face-name pair in a single loop; otherwise,
an additional reminder would be presented aer three iterations. is ensures that participants were familiarised
with each face. On each trial in the test phase, a question was presented (e.g. “Which is John’s nose?), followed by
a choice of two alternative images presented on the le and right sides of the screen, both horizontally centred. In
the part condition, the display consisted of two isolated features (two eyes, two noses, or two mouths), one was
from the target face, and the other was from the distractor face. In the whole condition, the display contained
two whole faces, with the target and a distractor face diering only with respect to one face part. Participants
were required to indicate if the target stimulus was on the le or on the right. e image pair remained on the
screen until response.
Stimuli were matched between the two conditions, such that facial parts tested in the part condition were also
tested in the whole condition. e whole and part conditions were randomly intermixed. Each block consisted
of 24 part and 24 whole trials. e order of block presentation was counterbalanced across participants.
Composite task. Faces were generated from 60 images (20 for each race; half females) of Chinese, Caucasian,
and African faces. Each face image was divided into two halves horizontally across the middle of the nose using
Adobe Photoshop CS6. e top and bottom halves from same-gender faces of dierent individuals were then
recombined at random, leaving a 3-pixel gap between the two parts. e top half and bottom halves were pre-
sented either aligned or misaligned (see Fig.4a). In the misaligned trials, the top and bottom face parts were
misaligned by shiing the top half horizontally to the le by half a face width. e same composite faces were
used in both conditions. is resulted in 40 aligned and 40 misaligned composite faces in total for each race
category. Stimuli in the aligned condition were 7.5° horizontal by 10.5° vertical while stimuli in the misaligned
condition were 11. 2° horizontal by 10.5° vertical.
Following Gauthier and Bukach17 (Fig.4a), in congruent trials, the top and bottom parts of the face were
created either from the same faces or from dierent faces (i.e., top-same and bottom-same or top-dierent and
bottom-dierent). On the other hand, in incongruent trials, one of the face halves was created from the same
face, while the other half was created from dierent faces (i.e., top-same and bottom-dierent or top-dierent
and bottom-same). is paradigm allows the calculation of a bias-free measure of sensitivity—d′ prime51,52.
Each trial started with a central xation cross for 500ms, followed by a centred face for 200ms. Aer a
Gaussian noise mask of 500ms, a test face appeared randomly in one of eight locations, each placed 1.2° from
the screen’s centre, for 200ms. Next, a blank screen was presented until a response was made. e participants’
task was to judge as quickly and accurately as possible whether the top half of the test face was identical to the
preceding study face while ignoring the task-irrelevant bottom half. ey were instructed to indicate their deci-
sion by pressing two keys on a keyboard (see Fig.4b). On each trial, both faces within a pair were either aligned
or misaligned, and these two conditions were intermixed. Trials were blocked by face race, and the order of
blocks was counterbalanced across participants. Hence, each participant performed three experimental blocks
of 80 trials (40 aligned and 40 misaligned), half of which consisted of face pairs that shared an identical top half
(same trials), and half of which consisted of face pairs with dierent top halves (dierent trials). Order of trial
presentation was fully randomised across participants. Participants rst completed 12 practice trials to ensure
that they understood the task.
Figure2. Experimental procedure for the learning and recognition stages in the yes–no recognition task.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Vol.:(0123456789)
Scientic Reports | (2021) 11:8507 |
www.nature.com/scientificreports/
Global–local task. is task is a variant of Navon’s53 task used in Wang etal.31 and assesses participants’ bias to
attend to the global shapes versus local shapes, or vice versa. In congruent shapes, the global and the local objects
forming the shapes shared an identity (e.g., local squares forming a global square). In incongruent shapes, the
shapes at the two levels had dierent identities (e.g., local circles forming a global square). In addition to congru-
ent and incongruent conditions, we also included a neutral (baseline) condition at both global and local levels in
which a task-irrelevant object (an X) forms the global or local shapes (see Fig.5). e Navon stimuli consisted
of shapes (circle, square or cross) with white outline presented on a black background. Each local shape was
0.5° × 0.5°; the local shapes were arranged to form a global square (4.9° × 4.9°), global circle (5.6° × 5.6°), or a
global cross (4.9° horizontal × 5.3° vertical).
ere were two blocks of trials, each containing 18 practice and 108 test trials. Each block was preceded by
instructions to identify the target shapes (circle and square) at either the global or local level as quickly and
accurately as possible. In each block, there were 36 congruent trials, 36 incongruent trials and 36 neutral trials
(18 local, 18 global). e neutral trials were included to serve as a baseline measure. e three main types of
trials were randomly intermixed. Each trial began with a blank screen (500ms), followed by a central xation
cross (700ms), en, a shape stimulus appeared randomly in one of the eight possible locations (0.49° away
from the centre of the screen) for 150ms, followed by a mask (48 × 48 array of diamonds each 0.19° × 0.19°) for
500ms. Participants were asked to indicate whether the target shape they saw was a circle or a square as fast as
possible. is task took approximately 3min. Each participant completed 216 trials in total (108 local-level and
108 global-level), with 18 practice trials in each block.
Figure3. Example of the stimuli of three dierent races used in the whole-part task. e whole-part eect
(WPE) relies on the assumption that it is much easier to identify the eyes (A), nose (B), or mouth (C) of the
target face when the features are shown in the context of the whole face than when they are shown in isolation.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Vol:.(1234567890)
Scientic Reports | (2021) 11:8507 |
www.nature.com/scientificreports/
Results
Distributions were normal as indicated by Kolmogorov–Smirnov test (all ps ≥ 0.1). e assumptions of homo-
geneity of variance were met in the three main measures (i.e., d′, accuracy, and mean response time) and no
violations were detected (Levene’s test all p > 0.05). Prior to each analysis for these three measures, outliers
further than two standard deviations from the mean were removed. For each ANOVA, Greenhouse–Geisser
corrections were applied whenever sphericity was violated. Follow-up tests were conducted using post-hoc tests
with Bonferroni correction for signicant main eects and planned comparisons for signicant interaction
eects. Bonferroni-corrected p values were reported. To ensure there was no speed-accuracy trade o, analyses
on face task performance were repeated using mean response times (RTs) as the dependent variable. Given that
the pattern of results was similar in the accuracy and RT data, in the interest of brevity, we report the response
time results in Supplementary Text.
Figure4. (a) Examples of the experimental design and (b) a sample of a ‘dierent’ trial used in composite-face
task. e participants’ task was to match the sequentially presented top halves while ignoring the bottom halves.
Figure5. Example of Navon stimuli for global–local task.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Vol.:(0123456789)
Scientic Reports | (2021) 11:8507 |
www.nature.com/scientificreports/
It is frequently argued that support for the null hypothesis being true cannot be obtained from the fact that
the p-values are larger than the alpha level (e.g.54–56). us, in addition to reporting the traditional null hypothesis
signicance tests, we also performed Bayesian analyses57,58 using the statistical soware JASP59 (0.14.0.0, https://
jasp- stats. org/) and the JASP default prior60,61 (Cauchy prior, r = 0.707; JASP Team, 2020). Bayesian analysis has
the pragmatic benet that it is not based on the evaluation of signicance levels that can be interpreted incor-
rectly, particularly when the results are non-signicant62. e Bayes Factor (BF10) provides the likelihood ratio
of the probability of the data given the alternative hypothesis (H1) divided by the probability of the same data
given the null hypothesis (H0). A BF10 value between 1 and 3 provides anecdotal evidence for H1; a value between
3 and 10 provides moderate evidence for H1; a value above 10 provides strong evidence for H1; a value between
1 and 1/3 provides anecdotal evidence for H0; a value between 1/10 and 1/3 provides moderate evidence for H0
and; a value less than 1/10 provides strong evidence for H0.
Yes–no recognition task. d-prime (d′) was used as an index of participants’ face recognition sensitivity.
In all cases where hit rate or false alarm rate equals 1.0, Snodgrass and Corwin’s63 correction was applied to
overcome innite values of d′. e d′ scores were then calculated by subtracting each participant’s z-score for
false-alarm rates from z-score for hit rates (d’ = ZH − ZFA )64. A two-way repeated measure analysis of variance
(ANOVA) was performed on d′, with face race (Chinese, Caucasian, and African) as within-subjects factor
and participant race (Malaysian-Chinese, European-Caucasian, African, and Australian-Caucasian) as between-
subjects factor.
Recognition accuracy (d′). Results from the ANOVA revealed a signicant main eect of Face Race, F (2,
230) = 24.14, p < 0.001, ηp2 = 0.17, BF10 = 1.25 × 106, but no main eect of Race of Observer was found, F (2,
87) = 1.80, p = 0.15, ηp2 = 0.05, BF10 = 0.23. Participants generally had highest recognition performance for Cauca-
sian faces, followed by African faces, and then Chinese faces (all ps < 0.05, BF10 ≥ 315.58). ere was a signicant
Face Race × Race of Observer interaction, F (6, 230) = 8.06, p < 0.001, ηp2 = 0.20, BF10 = 2.69 × 105 (see Fig.6). Pair-
wise comparisons (with p values Bonferroni corrected for multiple comparisons) conrmed that participants of
each of the ethnicities manifested an own-race recognition advantage. Malaysian Chinese were better at recog-
nising own-race faces than African faces (p = 0.02, BF10 = 15.55), but no dierence was found between own-race
and Caucasian faces (p = 0.95, BF10 = 0.25). European-Caucasians showed higher recognition sensitivity towards
own-race faces relative to Chinese and African faces (both p ≤ 0.001; BF10 = 41.75 and BF10 = 17.69, respectively).
Africans performed better for African and Caucasian faces than for Chinese faces (both p < 0.001; BF10 = 411.74
and BF10 = 1526.83, respectively) while no dierence was detected between African and Caucasian faces (p = 1,
BF10 = 0.20). Australian-Caucasians recognised own-race faces better than Chinese (p < 0.001, BF10 = 312.83) and
African faces (p = 0.008, BF10 = 8.60).
Whole‑part task. e whole part eect (WPE)—an index of holistic face processing—was calcuther-race
faces by using the formula lated by subtracting accuracy scores for part trials from those for whole trials. To con-
trol for any dierences in baseline accuracy, we computed the standardized WPE scores for own- and obelow65:
WPE
=
(%correct whole
−
%correctpart)
(%correct whole
+
%correct part).
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
Malaysian-Chinese Australian-Caucasian AfricanEuropean-Caucasian
Recognion Sensivity (d')
Race of Parcipant
Yes-No Recognion Ta sk
Chinese faces Caucasian faces African faces
**
*
**
**
**
**
**
Figure6. d’′ scores for the yes–no face recognition test of own- and other-race faces in Malaysian-Chinese,
Australian-Caucasian, African, and European-Caucasian participants. Error bars represent standard errors of
the mean (**p < 0.01; *p < 0.05).
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Vol:.(1234567890)
Scientic Reports | (2021) 11:8507 |
www.nature.com/scientificreports/
Whole‑part eect (WPE). A mixed ANOVA was performed on the magnitude of WPE, with Face Race as
within-subjects factor whereas Race of Participant as between-subjects factor. e main eect of Face Race was
signicant, F(2, 234) = 17.61, p < 0.001, ηp2 = 0.13, BF10 = 6.29 × 104, such that WPE was stronger for Chinese faces
than African (p < 0.001, BF10 = 4758.91) and Caucasian faces (p = 0.002, BF10 = 1201.85) while no dierence was
found between African and Caucasian faces (p = 1, BF10 = 0.13). Neither the main eect of Race of Participant,
F(3, 117) = 0.11, p = 0.95, ηp2 = 0.003, BF10 = 0.046, nor the critical two-way interaction between Face Race and
Race of Participant was signicant (see Fig.7), F(5.74, 223.86) = 1.21, p = 0.30, ηp2 = 0.03, BF10 = 0.019, suggest-
ing that the magnitude of WPE was not stronger for own-race faces than for other-race faces (Supplementary
TableS1). Complementary one-sample t tests split by participant race were computed to assess whether the mean
WPE scores were signicantly positive. Results conrmed that in each race group, the WPE scores were signi-
cantly greater than zero, not only for own-race faces, but also for other-race faces (all ps < 0.01, BF10 ≥ 19.12),
indicating the emergence of holistic face processing regardless of the dierent races of faces.
Composite face task. Holistic processing in the composite-face task was indicated by the performance
dierences between the congruent trials and incongruent trials. To further determine whether there was a dif-
ference in holistic face processing between own- and other-race faces within each race group, we then computed
the composite-face eect (CFE) score for each race of faces separately using the following formula66:
e magnitude of CFE between race groups was then examined with a mixed ANOVA, involving Face Race
as within-subjects variable and Race of Participant as between-subjects variable.
Composite face eect (CFE). A 3 (Face Race) by 4 (Race of Participant) ANOVA performed on the CFE scores
showed that neither the main eect of Race of Participant nor the main eect of Face Race was signicant, F(3,
117) = 0.44, p = 0.72, ηp2 = 0.01, BF10 = 0.027 and F(2,234) = 0.14, p = 0.87, ηp2 = 0.001, BF10 = 0.034, respectively.
No crossover interaction between Race of Participant and Face Race was found (see Fig.8), F(6, 234) = 0.91,
p = 0.49, ηp2 = 0.02, BF10 = 0.058, indicating similar holistic processing for both own- and other-race faces in each
race group (Supplementary TableS2). Complementary one-sample t-tests split by participant race showed that,
in most cases, the CFE scores were signicantly greater than zero, not only for own-race faces, but also for other-
race faces (all ps < 0.05, BF10 ≥ 2.27 × 103). e only exceptions were the CFEs for Caucasian faces in African par-
ticipants, t (29) = 1.19, p = 0.24, and for Chinese faces in European-Caucasian participants, t (29) = 1.17, p = 0.25.
Global–local task. Participants’ accuracy was near ceiling across trial types (mostly above 90%). is near-
perfect performance could potentially mask the global–local interference eect and render the results less reli-
able. erefore, our subsequent analyses focus on the response time (RT) instead to calculate the global–local
interference (GLI) scores, as traditionally done (e.g.31,53,67). Only RTs for correct responses were included in the
analysis and RTs for a trial were discarded if they were shorter than 200ms or longer than 2000ms. Preliminary
analysis on RTs showed that participants made slowest responses in incongruent trials (M = 536ms), followed
by the neutral trials (M = 520ms), and then the congruent trials (M = 503ms) (all p < 0.001), with neutral being
faster than congruent trials (p = 0.01), suggesting that neutral trials can serve as a baseline measure. Since per-
formance (both accuracy and RT) was not aected by whether the participants were tested on neutral-local
Congruency effect
=
d′congruent trials
−
d′incongruent trials,
Composite
−face effect (CFE)=congruency effect
aligned trials −misaligned trials
.
.00
.05
.10
.15
Malaysian-ChineseAustralian-CaucasianAfrican European-Caucasian
Whole-Part Effect
Race of Parcipant
Whole-Part Task
Chinese facesCaucasian faces African faces
Figure7. e magnitudes of the whole-part eect (WPE) for own- and other-race faces for each ethnic group
in whole-part face task. Error bars indicate standard errors of the mean.
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Vol.:(0123456789)
Scientic Reports | (2021) 11:8507 |
www.nature.com/scientificreports/
(mean accuracy = 93.42%, SD = 10.50%; mean RT = 537 ms, SD = 57ms) or neutral-global trials (mean accu-
racy = 97.02%, SD = 4.78%; mean RT = 530ms, SD = 63ms) (both ps > 0.05), we collapsed across these conditions
in the analysis.
To measure participants’ tendency to globally process general objects, a global–local interference (GLI) score
was calculated using the following formula for each participant by examining the degree to which global features
on the local incongruent trials interfere with RT.
Positive GLI scores indicate a global processing bias whereas negative GLI scores show a local processing bias.
GLI. As determined by one-way ANOVA, there was a statistically signicant dierence between race groups
(see Fig.9), F (3,117) = 10.81, ηp2 = 0.22, p < 0.001, BF10 = 53.32. Pairwise comparisons (with Bonferroni-cor-
rected p values) showed that the magnitude of GLI in Malaysian Chinese were signicantly greater than Austral-
ian Caucasians (p < 0.001, BF10 = 19.28) and marginally higher than European Caucasians (p = 0.09, BF10 = 0.30).
Similarly, Africans showed signicantly stronger GLI than European Caucasians (p = 0.04, BF10 = 1.03) and Aus-
tralian-Caucasians (p < 0.001, BF10 = 53.70). No signicant dierence was found between Malaysian Chinese and
Africans (p = 0.65, BF10 = 0.28), or between European- and Australian-Caucasians (p = 0.25, BF10 = 1.32).
GLI
=Congruent
global −local
−Incongruent(global −local)
Congruent
global +local
+Incongruent(global +local)
.
0.0
0.2
0.4
0.6
0.8
1.0
Malaysian-ChineseAustralian-CaucasianAfrican European-Caucasian
Magnitude of CFE (∆d')
Race of Parcipant
Composite Face Task
Chinese facesCaucasian facesAfrican faces
Figure8. e magnitudes of the composite face eect (CFE) for own- and other-race faces for each ethnic
group in composite face task. e CFE was calculated by subtracting congruency eect observed in misaligned
condition from that observed in aligned condition (i.e., the alignment by congruency interaction). Error bars
indicate ± 1 standard error of the mean.
.00
.01
.02
.03
.04
Malaysian-ChineseEuropean-CaucasianAfrican Australian-Caucasian
Global-to-Local Interference
Race of Parcipant
Global-Local Task
**
*
**
Figure9. e magnitude of global–local interference (GLI) as a function of participant group. Error bars
indicate standard errors of the mean. Asterisks indicate signicant dierences between race groups (**p < 0.01;
*p < 0.05).
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Vol:.(1234567890)
Scientic Reports | (2021) 11:8507 |
www.nature.com/scientificreports/
Correlation analyses. Pearson’s correlation analyses were performed to determine whether the face recog-
nition ability (FRA) for own- versus other-race faces was related to the three holistic processing indices: com-
posite-face eect (CFE), whole-part eect (WPE), and global–local interference (GLI). Rather than completely
excluding outliers with many valid observations from the inter-task correlational analyses, cases identied more
than 2 SDs from the mean for a particular measure were replaced by a score plus two times the standard devia-
tions. On this basis, less than 2% of the data were replaced within each task (yes–no task: 1.38%; whole-part task:
1.93%; composite-face task: 1.1%; global–local task: 0.83%). Aer Bonferroni-correct for multiple comparisons,
none of the correlations between FRA and measures of holistic processing (Table1) and between the ORE of
FRA and the ORE of holistic processing (Table2) was statistically signicant, suggesting that strength of the ORE
in face recognition was not predicted by strength of the ORE in holistic processing. To further support these null
Table 1. Pearson correlations (and corresponding p-values) between the holistic processing indices—whole-
part eect (WPE), the composite-face eect (CFE), and global–local interference (GLI)—with face recognition
ability (d′) as a function of stimulus race in each race group. chi Chinese faces, sa South African faces, cau
Caucasian faces. α (two-tailed) = 0.002 (0.05/21).
Malaysian Chinese (N = 31) Africans (N = 30) Australian-Caucasians (N = 30) European-Caucasians (N = 30)
chi d′sa d′cau d′chi d′sa d′cau d′chi d′sa d′cau d′chi d′sa d′cau d′
WPE
Chinese
faces − 0.50
(0.04) − 0.20
(0.28) − 0.46
(0.01) − 0.32
(0.08) − 0.20
(0.29) − 0.38
(0.04) 0.15 (0.43) 0.08 (0.69) − 0.08
(0.69) − 0.05
(0.81) − 0.35
(0.06) − 0.09 (0.63)
African
faces − 0.14
(0.46) − 0.12
(0.54) − 0.27
(0.15) − 0.35
(0.06) − 0.40
(0.03) − 0.17
(0.38) − 0.03
(0.87) − 0.15
(0.42) − 0.29
(0.13) 0.01 (0.95) − 0.26
(0.17) − 0.01 (0.94)
Caucasian
faces 0.28 (0.12) − 0.01
(0.97) − 0.02
(0.92) − 0.06
(0.76) − 0.02
(0.92) 0.10 (0.60) − 0.12
(0.92) − 0.06
(0.74) 0.06 (0.76) − 0.17
(0.38) − 0.28
(0.14) − 0.03 (0.87)
CFE
Chinese
faces 0.14 (0.45) 0.24 (0.20) 0.15 (0.41) 0.08 (0.68) 0.30 (0.11) 0.05 (0.80) 0.10 (0.61) 0.03 (0.87) − 0.08
(0.69) − 0.22
(0.25) − 0.05
(0.78) − 0.23 (0.22)
African
faces 0.24 (0.20) − 0.04
(0.84) 0.15 (0.43) 0.07 (0.71) − 0.30
(0.11) − 0.37
(0.04) − 0.02
(0.90) 0.31 (0.09) − 0.16
(0.41) 0.15 (0.43) − 0.13
(0.49) − 0.16 (0.40)
Caucasian
faces 0.31 (0.08) 0.24 (0.20) 0.30 (0.10) − 0.15
(0.43) 0.11 (0.57) − 0.20
(0.30) 0.28 (0.14) 0.01 (0.97) 0.06 (0.74) 0.15 (0.42) 0.02 (0.91) 0.04 (0.83)
GLI − 0.02
(0.92) 0.15 (0.43) 0.02 (0.90) 0.34 (0.06) 0.07 (0.70) − 0.20
(0.28) 0.41 (0.02) 0.18 (0.34) − 0.02
(0.94) − 0.15
(0.42) 0.08 (0.69) − 0.04 (0.82)
Table 2. Summary of Pearson’s correlations (corresponding p-values and Bayes factors) between the ORE
of face recognition ability (FRA), the OREs of holistic processing (WPE and CFE), and GLI by the race of
observers. chi Chinese faces, sa South African faces, cau Caucasian faces.α (two-tailed) = 0.01 (0.05/4).
Race of observers Pearson’s r p BF10
Chinese (N = 31)
FRA_chi_cau−WPE_chi_cau − 0.35 0.08 0.93
FRA_chi_cau−CFE_chi_cau − 0.06 0.77 0.23
FRA_chi_sa −WPE_chi_sa − 0.22 0.25 0.66
FRA_chi_sa −CFE_chi_sa − 0.09 0.64 0.48
European-Caucasian (N = 30)
FRA_cau_chi–WPE_cau_chi 0.15 0.42 0.31
FRA_cau_chi–CFE_cau_chi − 0.07 0.71 0.24
FRA_cau_sa–WPE_cau_sa 0.02 0.93 0.23
FRA_cau_sa–CFE_cau_sa 0.06 0.76 0.24
African (N = 30)
FRA_sa_chi–WPE_sa_chi − 0.14 0.47 0.29
FRA_sa_chi–CFE_sa_chi − 0.38 0.12 0.96
FRA_sa_cau–WPE_sa_cau − 0.11 0.56 0.27
FRA_sa_cau–CFE_sa_cau − 0.22 0.25 0.43
Australian-Caucasian (N = 30)
FRA_cau_chi–WPE_cau_chi 0.19 0.31 0.37
FRA_cau_chi–CFE_cau_chi − 0.05 0.78 0.24
FRA_cau_sa–WPE_cau_sa 0.21 0.27 0.41
FRA_cau_sa–CFE_cau_sa 0.30 0.11 0.75
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Vol.:(0123456789)
Scientic Reports | (2021) 11:8507 |
www.nature.com/scientificreports/
ndings, we performed the corresponding Bayesian correlation tests (Table2); for the ease of data visualisation,
the scatterplots were created (Supplementary Figs.S3–S6).
Discussion
is cross-cultural study aimed to systematically examine the relationship between holistic processing and
recognition of own- and other-race faces, by using Malaysian Chinese, African, European-Caucasian, and
Australian-Caucasian participants. e current experiment yielded four main results. First, the ORE for recog-
nition performance was pronounced in the face recognition task. Second, participants across race groups did
not show stronger holistic processing—as indexed by both the composite-face eect (CFE) and the whole-part
eect (WPE)—for own- than other-race faces. ird, in a global–local task, both Malaysian Chinese and Afri-
can participants were more susceptible to the GLI, indicating a stronger global processing bias as compared to
European- and Australian-Caucasian participants. Fourth, the WPE, the CFE, and the GLI were not associated
with face recognition performance for other-race faces, indicating that the ORE cannot be accounted for by
reduced face processing in global/holistic manner for other-race faces.
Across four race groups, participants exhibited a robust ORE in face recognition memory, although less
prominently for Caucasian faces. Most interestingly, Malaysian Chinese participants, who had grown up in a
highly multi-ethnic and Western-inuenced Asian country, performed equally well at recognising Chinese and
Caucasian faces, but less well at recognising African faces. is is consistent with the ndings by Wong etal.48
and Tan etal.28 (but see27). e latter study further explained the observed decit in the recognition of African
faces as a product of insucient visual experience, which leads to a core lack of perceptual ability in the face
system to extract the most diagnostic information from that face race. On the other hand, African participants
recognised African faces as well as they recognised Caucasian faces but were less good at recognising Chinese
faces. In contrast, both European- and Australian-Caucasian participants recognised Caucasian faces better
than Chinese and African faces.
Considering the relatively high proportion of ethnic Chinese people in Malaysia (42.3% in the Kuala Lum-
pur)68, we initially anticipated that Africans and European-Caucasian participants, who had resided in the
country for half a year or more on average prior to participating in this study, would recognise Chinese faces
well. However, this was not the case. e results showed that both African and European-Caucasian exchange/
transfer students were generally poor at recognising Chinese faces, indicating that staying in a multiracial envi-
ronment for a short period of time does not necessarily allow them to develop sensitivity to facial features that
are essential for recognising unfamiliar other-race faces. Given the reduced plasticity for face recognition in
adulthood69, a reduction of ORE would require sucient individuating experience during childhood69 and/or
explicit training70, rather than mere exposure to other-race faces71.
Malaysian Chinese and African participants were able to recognise Caucasian faces equally as well as their
own-race faces. ese results should not be too surprising, as Malaysian Chinese and African participants, who
were students attending a branch campus of a British university, were more likely to have increased exposure to
Caucasian faces in the mass media (e.g., western movies). Such a heightened experience in actively individuating
them in everyday life might lead to improvements in perceptual sensitivity to diagnostic features on Caucasian
faces.
To test the holistic account of the ORE, we used two direct (but uncorrelated37) measures of holistic process-
ing: the composite-face and whole-part tasks. In both measures, we did not nd evidence of stronger holistic pro-
cessing eect for own- than other-race faces. is eect is remarkable because it was consistent across all our race
groups. Although a few studies have found stronger holistic processing for own compared to other race faces11,65,
these results are not always replicated. In fact, considerable evidence has accumulated suggesting that holistic
processing occurs for other race faces23,24, for facial morphologies that are less visually experienced13,38–40, and
even for other-species faces72. Our results thus run counter to the prediction derived from the holistic account
of ORE that the magnitude of holistic processing would be stronger for own-race faces than for other-race faces.
It is tempting to interpret our results as showing that the holistic processing for own- and other-race faces is
comparable in magnitude. To seek evidence that support the null hypothesis, we additionally performed Bayesian
statistical analysis for two lines of results: (a) the magnitudes of holistic processing are not stronger for own- than
other-race faces (see Supplementary TableS4); and (b) neither the CFE or WPE are highly correlated with the
face recognition performance. e results are summarised in Tables1 and 2, where the overall pattern of results
is consistent with those obtained via NHST (null hypothesis signicance testing) analysis. However, one caveat
is that, aer adjusting for multiple comparisons in the NHST analyses, there were a few cases of a weak, non-
signicant pattern of stronger holistic eects for own-race or specic-race faces (e.g., there were suggestions of
a stronger WPE eect for Chinese and African participants looking at Chinese faces), and so caution should be
exercised in drawing this conclusion based on null ndings. In addition, despite a very large sample size rela-
tive to prior work and a pronounced ORE, in terms of accuracy, for the composite-face and whole-part tasks,
these measures may not have been suciently sensitive to capture racial dierences in holistic processing even
at standard experimental sizes. us, the interpretation of CFE and WPE data must also be taken with caution
unless they can be replicated with a larger sample size.
Holistic processing has been found to be associated with face recognition performance31,73 and the ORE
magnitude16. In the present study, however, participants’ memory for own- and other-race faces did not seem
to be aected by the magnitude of holistic processing. e failure to nd evidence for a correlation is surpris-
ing given the dominant theme in the literature that holistic processing is important for both perceiving and
recognising faces. is null nding cannot be attributed to any confound derived from the stimulus variability
because observers of dierent races were always better recognising own-race faces (i.e. ORE) across face tasks
(see Supplementary Figs.S1, S2).
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Vol:.(1234567890)
Scientic Reports | (2021) 11:8507 |
www.nature.com/scientificreports/
Rather, it suggests that holistic processing, which lacks reliable individual dierences74, is not directly associ-
ated with dierences in recognition memory performance for own- and other-race faces. Extensive individuat-
ing experience with own-race faces could enhance face recognition ability75, but such experience may not be
required to generalise holistic processing to other races of faces. Such an interpretation is consistent with the
idea that holistic processing for other-race faces can be easily employed without being restricted by an intrinsic,
context-dependent capacity76.
Publication bias is a possible explanation when an eect does not replicate77. It is relatively easy to publish
results showing a dierence between two groups, even if the dierence was unpredictable, small and hard to
explain. It is likely that the published papers overstate the dierences in holistic processing between own- and
other-race faces. Our current results resonate with several recent studies showing that holistic processing is not
directly linked with face recognition ability18,24 and can be elicited by both own- and other-race faces without
extensive individuating experience38. Taken together, these observations challenge the assertion that the ORE in
face recognition is a consequence of reduced holistic processing for other-race faces. Holistic processing may play
a signicant role in the early stages of face recognition78, possibly at the level of face detection or face matching
that place lower cognitive demands on memory; however, it is not sucient for explaining the dierences in
recognition for own- and other-race faces. is rather varied evidence also indicates that the degree of holistic
processing applied to a face stimulus may not be as strongly modulated by its perceived race identity as commonly
expected; instead, it seemed to be somewhat dependent on the facial physiognomy, stimulus characteristics and
tasks performed on them79,80.
Overall, our results suggest that, regardless of the race, faces are processed holistically and that there is no
strong association between holistic processing and recognition of own and other race faces. ese ndings have
an important theoretical implication, namely that holistic processing is necessary but not sucient for face
identication81,82. Although holistic processing would allow the fast binding of facial features into a coherent
global percept, this representation would need then to be further processed by a specialised face recognition
mechanism83. In the same vein, our results support the notion that the origins of holistic face processing are
better accounted for by the template hypothesis rather than the attentional strategy hypothesis (for reviews,
see4). While the attention strategy hypothesis proposes that holistic processing—a strategy of attending to all
face parts simultaneously—is shaped by the experience from frequent social interactions and regular exposure to
faces4,19, the template hypothesis postulates that faces are represented as a single unit to t a memory template6,84
which may be established innately85. Our current results that holistic processing can be elicited by both own-
and other-race faces without extensive individuating experience seem more consistent with holistic processing
being a consequence of the representational constraints of a global face template rather than the inexibility in
attentional weightings on face parts.
Another open question is whether people possess the necessary perceptual abilities to recognise other-race
faces at the level of the individual, but only lack the social motivation to do so86. According to the social-cognitive
position, the source of the ORE is not perceptual, but a resistance to individuate other-race faces due to their out-
group status. Hence, the emergence of the ORE may be due to motivational factors rather changes in perceptual
expertise. Alternatively, ORE could be a product of converging factors involving social categorization, motivated
individuation, and perceptual experience; for example, neither raw perceptual exposure nor the motivation to
individuate is sucient to attenuate the ORE but requires both the proper motivation and practice to individuate
other-race faces. Further research is required to conrm these hypotheses.
Here we also provide the rst study to use Navon gures to compare global–local processing dierences
between Malaysian Chinese, African, Australian Caucasian, and European Caucasian participants. Our results
show that both Malaysian Chinese and African groups were more susceptible to global–local interference (GLI)
than Caucasian groups (European and Australian), indicating a reduced ability to inhibit the inuence of holistic
information on piecemeal processing. Not only is this result in agreement with numerous studies that provided
evidence of stronger global processing in collectivist societies (i.e. the East), and weaker local processing, as
compared to individualistic societies (e.g. the West)41,87, but also the rst report that Africans showed a global
processing bias stronger than that of Westerners. is lends strong empirical support to the notion that infor-
mation-gathering strategy (global versus local processing) for general stimuli can be culture-dependent25,42.
Furthermore, in line with the domain-specicity hypothesis, the magnitude of GLI did not signicantly correlate
with holistic face processing measures and face recognition ability, implying that such low-level perceptual biases
for information processing may not necessarily be generalizable to high-level face processing tasks.
In conclusion, the current study did not nd evidence that holistic processing was stronger for own- than
other-race faces. Interestingly, holistic processing for other-race faces did not preclude the observation of OREs.
e current ndings not only contrast with the assumptions that holistic processing is stronger for own-race faces,
but also question the commonly claimed evidence in support of a strong association between face memory and
holistic face processing. ese results converge with recent studies questioning the holistic processing account of
the ORE. Future research is needed to help elucidate the fundamental roles of cognitive and perceptual orienting
mechanisms, other than holistic processing, that may underlie the recognition of own- and other-race faces.
Received: 21 June 2020; Accepted: 31 March 2021
References
1. Meissner, C. A. & Brigham, J. C. irty years of investigating the own-race bias in memory for faces: A meta-analytic review.
Psychol. Public Policy Law 7(1), 3–35. https:// doi. org/ 10. 1037// 1076- 8971.7. 1.3 (2001).
2. Platz, S. J. & Hosch, H. M. Cross-racial/ethnic eyewitness identication: A eld study. J. Appl. Soc. Psychol. 18(11), 972–984 (1988).
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Vol.:(0123456789)
Scientic Reports | (2021) 11:8507 |
www.nature.com/scientificreports/
3. Rossion, B. & Michel, C. An experience-based holistic account of the other-race face eect. In e Oxford Handbook of Face
Perception (eds Calder, A. et al.) 215–244 (Oxford University Press, 2011).
4. Maurer, D., Le Grand, R. & Mondloch, C. J. e many faces of congural processing. Trends Cogn. Sci. 6(6), 255–260. https:// doi.
org/ 10. 1016/ S1364- 6613(02) 01903-4 (2002).
5. Rossion, B. e composite face illusion: A whole window into our understanding of holistic face perception. Vis. Cogn. 21(2),
1–115. https:// doi. org/ 10. 1080/ 13506 285. 2013. 772929 (2013).
6. Tanaka, J. W. & Farah, M. J. Parts and wholes in face recognition. Q. J. Exp. Psychol. 46(2), 225–245. https:// doi. org/ 10. 1080/ 14640
74930 84010 45 (1993).
7. Tanaka, J. W., Simonyi, D., Tanaka, J. W. & Simonyi, D. e “parts and wholes” of face recognition: A review of the literature. Q. J.
Exp. Psychol. 69(10), 1876–1889. https:// doi. org/ 10. 1080/ 17470 218. 2016. 11467 80 (2016).
8. Hayward, W. G., Crookes, K. & Rhodes, G. e other-race eect: Holistic coding dierences and beyond. Vis. Cogn. 21(9–10),
1224–1247. https:// doi. org/ 10. 1080/ 13506 285. 2013. 824530 (2013).
9. Tanaka, J. W., Kiefer, M. & Bukach, C. M. A holistic account of the own-race eect in face recognition: Evidence from a cross-
cultural study. Cognition 93(1), 1–9. https:// doi. org/ 10. 1016/j. cogni tion. 2003. 09. 011 (2004).
10. Michel, C., Rossion, B., Han, J., Chung, C.-S. & Caldara, R. Holistic processing is nely tuned for faces of one’s own race. Psychol.
Sci. 17(7), 608–615. https:// doi. org/ 10. 1111/j. 1467- 9280. 2006. 01752.x (2006).
11. Bukach, C. M., Cottle, J., Ubiwa, J. & Miller, J. Individuation experience predicts other-race eects in holistic processing for both
Caucasian and Black participants. Cognition 123(2), 319–324. https:// doi. org/ 10. 1016/j. cogni tion. 2012. 02. 007 (2012).
12. Crookes, K., Favelle, S. & Hayward, W. G. Holistic processing for other-race faces in Chinese participants occurs for upright but
not inverted faces. Front. Psychol. 4(29), 1–9. https:// doi. org/ 10. 3389/ fpsyg. 2013. 00029 (2013).
13. Mondloch, C. J. et al. Processes underlying the cross-race eect: An investigation of holistic, featural, and relational processing of
own-race versus other-race faces. Perception 39(8), 1065–1085. https:// doi. org/ 10. 1068/ p6608 (2010).
14. Ross, D., Richler, J. J. & Gauthier, I. Reliability of composite-task measurements of holistic face processing. Behav. Res. Methods
47, 736–743. https:// doi. org/ 10. 3758/ s13428- 014- 0497-4 (2014).
15. Ross, D. & Gauthier, I. Holistic processing of faces in the composite task depends on size. J. Vis. 14(10), 571–571. https:// doi. org/
10. 1167/ 14. 10. 571 (2014).
16. DeGutis, J., Mercado, R. J., Wilmer, J. & Rosenblatt, A. Individual dierences in holistic processing predict the own-race advantage
in recognition memory. PLoS ONE 8(4), 1–13. https:// doi. org/ 10. 1371/ journ al. pone. 00582 53 (2013).
17. Gauthier, I. & Bukach, C. Should we reject the expertise hypothesis? Cognition 103(2), 322–330. https:// doi. org/ 10. 1016/j. cogni
tion. 2006. 05. 003 (2007).
18. Richler, J. J., Cheung, O. S. & Gauthier, I. Holistic processing predicts face recognition. Psychol. Sci. 22(4), 464–471. https:// doi.
org/ 10. 1177/ 09567 97611 401753 (2011).
19. Richler, J. J. & Gauthier, I. A meta-analysis and review of holistic face processing. Psychol. Bull. 140(5), 1281–1302. https:// doi.
org/ 10. 1037/ a0037 004.A (2014).
20 . Hills, P. J., Ross, D. A. & Lewis, M. B. Attention misplaced: e role of diagnostic features in the face-inversion eect. J. Exp. Psychol.
Hum. Percept. Perform. 37(5), 1396–1406. https:// doi. org/ 10. 1037/ a0024 247 (2011).
21. Rhodes, G. et al. Features, conguration, and holistic face processing. In Oxford Handbook of Face Perception (eds Rhodes, G. et
al.) (Oxford University Press, Oxford, 2011).
22. Hancock, K. J. & Rhodes, G. Contact, congural coding and the other-race eect in face recognition. Br. J. Psychol. 99, 45–56.
https:// doi. org/ 10. 1348/ 00071 2607X 199981 (2008).
23. Richler, J. J., Wong, Y. K. & Gauthier, I. Perceptual expertise as a shi from strategic interference to automatic holistic processing.
Curr. Dir. Psychol. Sci. 20(2), 129–134. https:// doi. org/ 10. 1177/ 09637 21411 402472 (2011).
24. Richler, J. J., Floyd, R. J. & Gauthier, I. About-face on face recognition ability and holistic processing. J. Vis. 15(9), 15. https:// doi.
org/ 10. 1167/ 15.9. 15 (2015).
25. McKone, E. et al. Asia has the global advantage: Race and visual attention. Vis. Res. 50(16), 1540–1549. https:// doi. org/ 10. 1016/j.
visres. 2010. 05. 010 (2010).
26 . Michel, C., Caldara, R. & Rossion, B. Same-race faces are perceived more holistically than other-race faces. Vis. Cogn. 14(1), 55–73.
https:// doi. org/ 10. 1167/4. 8. 425 (2006).
27. Estudillo, A. J., Lee, J. K. W., Mennie, N. & Burns, E. No evidence of other-race eect for Chinese faces in Malaysian non-Chinese
population. Appl. Cogn. Psychol. 34, 270–276. https:// doi. org/ 10. 1002/ acp. 3609 (2020).
28. Tan, C. B. Y., Stephen, I. D., Whitehead, R. & Sheppard, E. You look familiar: How Malaysian Chinese recognize faces. PLoS ONE
7(1), 1–4. https:// doi. org/ 10. 1371/ journ al. pone. 00297 14 (2012).
29. Gao, Z., Flevaris, A. V., Robertson, L. C. & Bentin, S. Priming global and local processing of composite faces: Revisiting the
processing-bias eect on face perception. Atten. Percept. Psychophys. 73(5), 1477–1486. https:// doi. org/ 10. 3758/ s13414- 011- 0109-7
(2011).
30. Gerlach, C. & Krumborg, J. R. Same, same—But dierent: On the use of Navon derived measures of global/local processing in
studies of face processing. Acta Physiol. (Oxf.) 153, 28–38. https:// doi. org/ 10. 1016/j. actpsy. 2014. 09. 004 (2014).
31. Wang, R., Li, J., Fang, H., Tian, M. & Liu, J. Individual dierences in holistic processing predict face recognition ability. Psychol.
Sci. 23(2), 169–177. https:// doi. org/ 10. 1177/ 09567 97611 420575 (2012).
32. Behrmann, M., Richler, J. J., Avidan, G. & Kimchi, R. Holistic face perception. In Oxford Handbook of Perceptual Organization (ed.
Wagemans, J.) 758–774 (Oxford University Press, 2015).
33. Duchaine, B. & Yovel, G. A revised neural framework for face processing. Annu. Rev. Vis. Sci. 1, 393–416. https:// doi. o rg/ 10. 1146/
annur ev- vision- 082114- 035518 (2015).
34. Piepers, D. W. & Robbins, R. A. A review and clarication of the terms “holistic”, “congural”, and “relational” in the face percep-
tion literature. Front. Psychol. 3, 1–11. https:// doi. org/ 10. 3389/ fpsyg. 2012. 00559 (2012).
35. Rhodes, G. Looking at faces: First-order and second-order features as determinants of facial appearance. Perception 42(11),
1179–1199. https:// doi. org/ 10. 1068/ p1700 43n (2013).
36. Tanaka, J. W. & Gordon, I. Features, conguration, and holistic face processing. In e Oxford Handbook of Face Perception (eds
Calder, A. J. et al.) 177–194 (Oxford University Press, 2011).
37. Rezlescu, C., Susilo, T., Wilmer, J. B. & Caramazza, A. e inversion, part-whole, and composite eects reect distinct perceptual
mechanisms with varied relationships to face recognition. J. Exp. Psychol. Hum. Percept. Perform. 43(12), 1961–1973. https:// doi.
org/ 10. 1037/ xhp00 00400 (2017).
38. Harrison, S. A., Gauthier, I., Hayward, W. G. & Richler, J. J. Other-race eects manifest in overall performance, not qualitative
processing style. Vis. Cogn. 22(6), 843–864. https:// doi. org/ 10. 1080/ 13506 285. 2014. 918912 (2014).
39. Hayward, W. G., Rhodes, G. & Schwaninger, A. An own-race advantage for components as well as congurations in face recogni-
tion. Cognition 106(2), 1017–1027. https:// doi. org/ 10. 1016/j. cogni tion. 2007. 04. 002 (2008).
40. Horry, R., Cheong, W. & Brewer, N. e other race eect in perception and recognition: Insights from the complete composite
task. J. Exp. Psychol. Hum. Percept. Perform. 41(2), 508–524. https:// doi. org/ 10. 1007/ s13398- 014- 0173-7.2 (2015).
41 . Masuda, T. & Nisbett, R. E. Attending holistically versus analytically: Comparing the context sensitivity of Japanese and Americans.
J. Pers. Soc. Psychol. 81(5), 922–934. https:// doi. org/ 10. 1037/ 0022- 3514. 81.5. 922 (2001).
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Vol:.(1234567890)
Scientic Reports | (2021) 11:8507 |
www.nature.com/scientificreports/
42. Lao, J., Vizioli, L. & Caldara, R. Culture modulates the temporal dynamics of global/local processing. Cult. Brain 1, 158–174.
https:// doi. org/ 10. 1007/ s40167- 013- 0012-2 (2013).
43. Chiao, J. Y. et al. Neural basis of individualistic and collectivistic views of self. Hum. Brain Mapp. 30(9), 2813–2820. https:// doi.
org/ 10. 1002/ hbm. 20707 (2009).
44. Eaton, L. & Louw, J. Culture and self in South Africa: Individualism–collectivism predictions. J. Soc. Psychol. 140(2), 210–217.
https:// doi. org/ 10. 1080/ 00224 54000 96004 61 (2000).
45. Faul, F., Erdfelder, E., Lang, A.-G. & Buchner, A. G*Power 3: A exible statistical power analysis for the social, behavioral, and
biomedical sciences. Behav. Res. Methods 39, 175–191 (2007).
46. Coetzee, V., Chen, J., Perrett, D. I. & Stephen, I. D. Deciphering faces: Quantiable visual cues to weight. Perception 39(1), 51–61.
https:// doi. org/ 10. 1068/ p6560 (2010).
47. Bainbridge, W. A., Isola, P. & Oliva, A. e intrinsic memorability of face photographs. J. Exp. Psychol. Gen. 142(4), 1323–1334.
https:// doi. org/ 10. 1037/ a0033 872 (2013).
48 . Wong, H. K., Stephen, I. D. & Keeble, D. R. T. e own-race bias for face recognition in a multiracial society. Front. Ps ychol. 11(208),
1–16. https:// doi. org/ 10. 3389/ fpsyg. 2020. 00208 (2020).
49. Chen, J. & Tiddeman, B. Multi-cue facial feature detection and tracking under various illuminations. Int. J. Robot. Autom. 25(2),
162–171 (2010).
50. Macrae, C. N. & Lewis, H. L. Do I know you? Processing orientation and face recognition. Psychol. Sci. 13(2), 194–196. https://
doi. org/ 10. 1111/ 1467- 9280. 00436 (2002).
51. Richler, J. J., Cheung, O. S. & Gauthier, I. Holistic processing predicts face recognition. Psychol. Sci. 22(4), 464–471. https:// doi.
org/ 10. 1177/ 09567 97611 401753 (2012).
52. Richler, J. J., Gauthier, I., Wenger, M. J. & Palmeri, T. J. Holistic processing of faces: Perceptual and decisional components. J. Exp.
Psychol. Learn. Mem. Cogn. 34(2), 328–342. https:// doi. org/ 10. 1037/ 0278- 7393. 34.2. 328 (2008).
53. Navon, D. Forest before the trees. the precedence of global features in visual perception. Cogn. Psychol. 9, 353–383 (1977).
54. Cohen, J. e earth is round (p < .05). Am. Psychol. 49(12), 997–1003. https:// doi. org/ 10. 1037/ 0003- 066X. 49. 12. 997 (1994).
55. Greenland, S. et al. Statistical tests, P values, condence intervals, and power: A guide to misinterpretations. Eur. J. Epidemiol.
31(4), 337–350. https:// doi. org/ 10. 1007/ s10654- 016- 0149-3 (2016).
56. Makin, T. R. & Orban de Xivry, J.-J. Ten common statistical mistakes to watch out for when writing or reviewing a manuscript.
Elife 8, 1–13. https:// doi. org/ 10. 7554/ eLife. 48175 (2019).
57. Dienes, Z. Bayesian versus orthodox statistics: Which side are you on? Perspect. Psychol. Sci. 6(3), 274–290. https:// doi. org/ 10.
1177/ 17456 91611 406920 (2011).
58. Rouder, J. N., Morey, R. D., Verhagen, J., Swagman, A. R. & Wagenmakers, E.-J. Bayesian analysis of factorial designs. Psychol.
Methods 22(2), 304–321. https:// doi. org/ 10. 1037/ met00 00057 (2017).
59. JASP Team (2020). JASP (Version 0.14.1) [Computer soware].
60. Wagenmakers, E.-J., van Ravenzwaaij, D, & de Ron, J. Concerns About the Default Cauchy are Oen Exaggerated: A Demonstration
with JASP 0.12. (2020) (accessed 14 May 2020); Bayesian Spectacles. https:// www. bayes iansp ectac les. org/ conce rns- a bou t- the- defau
lt- cauchy- are- oen- exagg erated- a- demon strat ion- with- jasp-0- 12/.
61. Jereys, H. eory of Probability 3rd edn. (Clarendon Press, 1961).
62. Rouder, J. N., Speckman, P. L., Sun, D., Morey, R. D. & Iverson, G. Bayesian t tests for accepting and rejecting the null hypothesis.
Psychon. Bull. Rev. 16(2), 225–237. https:// doi. org/ 10. 3758/ PBR. 16.2. 225 (2009).
63 . Snodgrass, J. G. & Corwin, J. Pragmatics of measuring recognition memory : Applications to dementia and amnesia. J. Exp. Psychol.
Gen. 117(1), 34–50. https:// doi. org/ 10. 1037/ 0096- 3445. 117.1. 34 (1988).
64. Macmillan, N. A. & Creelman, C. D. Detection eory: A User’s Guide (Cambridge University Press, 1991).
65. Zhu, Q., Li, X., Chow, K. & Liu, J. e part task of the part-spacing paradigm is not a pure measurement of part-based information
of faces. PLoS ONE 4(7), e6239. https:// doi. org/ 10. 1371/ journ al. pone. 00062 39 (2009).
66. de Heering, A. & Rossion, B. Prolonged visual experience in adulthood modulates holistic face perception. PLoS ONE 3(5), 1–5.
https:// doi. org/ 10. 1371/ journ al. pone. 00023 17 (2008).
67. Dale, G. & Arnell, K. M. Lost in the forest, stuck in the trees: Dispositional global/local bias is resistant to exposure to high and
low spatial frequencies. PLoS ONE 9(7), 14. https:// doi. org/ 10. 1371/ journ al. pone. 00986 25 (2014).
68. Department of Statistics Malaysia. e 2010 Population and Housing Census of Malaysia (2010) (accessed 28 Oct 2019); https://
www. dosm. gov. my/ v1/ index. php?r= column/ cthem eByCa t& cat= 117& bul_ id= MDMxd HZ jWT k1SjF zTzNk RXYzc VZjdz 09& menu_
id= L0phe U43NW JwRWV SZklW dzQ4T lhUUT 09.
69. Mckone, E. et al. A critical period for faces: Other-race face recognition is improved by childhood but not adult social contact.
Sci. Rep. https:// doi. org/ 10. 1038/ s41598- 019- 49202-0 (2019).
70 . Tanaka, J. W., Heptonstall, B. & Hagen, S. Perceptual expertise and the plasticity of other-race face recognition. Vis. Cogn. 21(9–10),
1183–1201. https:// doi. org/ 10. 1080/ 13506 285. 2013. 826315 (2013).
71. Walker, P. M. & Hewstone, M. A developmental investigation of other-race contact and the own-race face eect. Br. J. Dev. Psychol.
24(3), 451–463. https:// doi. org/ 10. 1348/ 02615 1005X 51239 (2006).
72. Taubert, J. Chimpanzee faces are ‘special’ to humans. Perception 38(3), 343–356. https:// doi. org/ 10. 1068/ p6254 (2009).
73. DeGutis, J., Wilmer, J., Mercado, R. J. & Cohan, S. Using regression to measure holistic face processing reveals a strong link with
face recognition ability. Cognition 126(1), 87–100. https:// doi. org/ 10. 1016/j. cogni tion. 2012. 09. 004 (2013).
74. Sunday, M. A., Richler, J. J. & Gauthier, I. Limited evidence of individual dierences in holistic processing in dierent versions of
the part-whole paradigm. Atten. Percept. Psychophys. https:// doi. org/ 10. 3758/ s13414- 017- 1311-z (2017).
75. Rhodes, G. et al. Contact and other-race eects in congural and component processing of faces. Br. J. Psychol. 100(4), 717–728.
https:// doi. org/ 10. 1348/ 00071 2608X 396503 (2009).
76. McKone, E., Brewer, J. L., MacPherson, S., Rhodes, G. & Hayward, W. G. Familiar other-race faces show normal holistic processing
and are robust to perceptual stress. Perception 36(2), 224–248. https:// doi. org/ 10. 1068/ p5499 (2007).
77. Francis, G. Publication bias and the failure of replication in experimental psychology. Psychon. Bull. Rev. 19(6), 975–991. https://
doi. org/ 10. 3758/ s13423- 012- 0322-y (2012).
78. Richler, J. J., Mack, M. L., Gauthier, I. & Palmeri, T. J. Holistic processing of faces happens at a glance. Vis. Res. 49(23), 2856–2861.
https:// doi. org/ 10. 1016/j. visres. 2009. 08. 025 (2009).
79. Wang, Z. et al. An other-race eect for congural and featural processing of faces: Upper and lower face regions play dierent
roles. Front. Psychol. 6(559), 1–8. https:// doi. org/ 10. 3389/ fpsyg. 2015. 00559 (2015).
80. Zhao, M., Bültho, H. H. & Bültho, I. A shape-based account for holistic face processing. J. Exp. Psychol. Learn. Mem. Cogn.
42(4), 584–597. https:// doi. org/ 10. 1037/ xlm00 00185 (2016).
81. Watson, T. L. Implications of holistic face processing in autism and schizophrenia. Front. Psychol. 4(414), 1–11. https:// doi. org/
10. 3389/ fpsyg. 2013. 00414 (2013).
82. Watson, T. L. & Robbins, R. A. e nature of holistic processing in face and object recognition: Current opinions. Front. Psychol.
5(3), 1–2. https:// doi. org/ 10. 3389/ fpsyg. 2014. 00003 (2014).
83. Taubert, J. & Alais, D. e composite illusion requires composite face stimuli to be biologically plausible. Vis. Res. 49(14), 1877–
1885. https:// doi. org/ 10. 1016/j. visres. 2009. 04. 025 (2009).
84. Farah, M. J., Wilson, K. D. & Tanaka, J. N. What is “special” about face perception? Psychol. Rev. 105(3), 482–498 (1998).
Content courtesy of Springer Nature, terms of use apply. Rights reserved
Vol.:(0123456789)
Scientic Reports | (2021) 11:8507 |
www.nature.com/scientificreports/
85. McKone, E. et al. Can generic expertise explain special processing for faces? Trends Cogn. Sci. 11(1), 8–15. https:// doi. org/ 10.
1016/J. TICS. 2006. 11. 002 (2007).
86. Hugenberg, K., Miller, J. & Claypool, H. M. Categorization and individuation in the cross-race recognition decit: Toward a solu-
tion to an insidious problem. J. Exp. Soc. Psychol. 43(2), 334 (2007).
87 . Nisbett, R. E. & Miyamoto, Y. e inuence of culture: Holistic versus analytic perception. Trends Cogn. Sci. 9(10), 467–473. https://
doi. org/ 10. 1016/j. tics. 2005. 08. 004 (2005).
Author contributions
H.K. and D.K. conceived the experiment. H.K. prepared the stimuli and conducted the experiment. H.K. and
I.S. analysed the results. H.K. and A.E. wrote the main manuscript text. All authors reviewed the manuscript.
Competing interests
e authors declare no competing interests.
Additional information
Supplementary Information e online version contains supplementary material available at https:// doi. org/
10. 1038/ s41598- 021- 87933-1.
Correspondence and requests for materials should be addressed to H.K.W.
Reprints and permissions information is available at www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional aliations.
Open Access is article is licensed under a Creative Commons Attribution 4.0 International
License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the
Creative Commons licence, and indicate if changes were made. e images or other third party material in this
article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons licence and your intended use is not
permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.
© e Author(s) 2021
Content courtesy of Springer Nature, terms of use apply. Rights reserved
1.
2.
3.
4.
5.
6.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:
use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at
onlineservice@springernature.com
Content uploaded by Hoo Keat Wong
Author content
All content in this area was uploaded by Hoo Keat Wong on Apr 19, 2021
Content may be subject to copyright.
